Genealogical and Genetic Evidence Relating to the Native American Ancestry of: Margaret Ann (Hensiek) Faux

Transcription

1 Genealogical and Genetic Evidence Relating to the Native American Ancestry of: Margaret Ann (Hensiek) Faux One of the profound difficulties in exploring the early genealogy of Ozark families is that there are simply few to no records to consult relating to certain key times and places. For example civil registration of births, marriages and deaths did not begin in Missouri until The useful census records do not emerge until 1850, and there are likely to be few if any church records to consult relating to a mobile frontier population. Often families will have Indian princess ancestor stories (most of which tend to be proved wrong or wildly distorted). It is common for Appalachian and Ozark families to have an oral tradition that some ancestor (not necessarily identified) was a full blooded then name your tribe. Often it seems that a family will claim Cherokee ancestry again, generally without what would be considered good genealogical evidence where one can use the preponderance of evidence litmus test. So, reality dictates that the criteria be amended slightly, however oral traditions or stories must have some paper trail support or they will always remain hypotheses. One solution is to seek out any pieces of documentary information, even if marginal, and then use genetic ancestry testing to validate the other findings. In some families, the approach will be to go from a finding of Native American ancestry suggested by one of the reliable and valid genetic tests, then move toward seeking documentary support. The latter applies to the family of Margaret Faux. We first became alerted to the possibility of Native American ancestry after Margaret s 100 year old mother, Flossie Fern (Arthur) Hensiek, obtained positive findings on a test called DNAPrint. Prior to outlining all of the genetic testing on Margaret and her mother, the author will first provide the only evidence, albeit circumstantial, of a possible link to the Shawnee Nation. This is in the form of the biography of John Arthur, the half brother of Flossie Fern s grandfather. GENEALOGICAL EVIDENCE Biographical Sketch of John Arthur, Dent County, Missouri "History of Laclede, Camden, Dallas, Webster, Wright, Texas, Pulaski, Phelps & Dent Counties, Missouri, Published 1889, Goodspeed Publishing Company. John Arthur, tiller of the soil and stock raiser of Watkin Township, was born in Washington (now Marion) County, KY., March 20, 1814, and is one of the old and enterprising citizens of the township. He is the son of Barnabas and Nancy (Vaughn) Arthur, the former probably a native of Virginia, and the latter a native of Washington County, KY. The mother died a few years after her marriage, and the father married again and removed to Boone's Lick, in North Missouri, where he spent one year, returned to Kentucky, and at the end of one year removed to Franklin County, MO., and again 1

2 returned to Kentucky, but later removed Crawford County, MO., where they both, father and step-mother, passed their last days. Mr. Arthur was one of the pioneers of South Missouri, settling there when it was a wilderness, and when it was probably a Territory. He reared four sons and a daughter, John Arthur being the only one living of the first family. He was reared principally where Meramec Iron Works were when the country was one vast wilderness, teeming with bear, wolves, panthers, elk, deer, etc., and it was common for fifteen years to see from 1,000 to 1,500 Shawnee Indians. He was reared with very little schooling, and when not working, he generally spent a great deal of time in hunting with the Indians, and learning the Shawnee language, in which he learned to converse readily. Before he was twelve years of age he had killed a bear. When old enough, he began working in the old Meramec Iron Works, where he was employed until twenty-four years of age. It is safe to say that without some sort of family connection, certainly in those days when the settlers and the Shawnee were unfriendly terms and entirely avoided each other if at all possible, the above statements would not make sense. Hence the information in this biography is taken to support a possible family connection either through the father Barnabus Arthur or the latter s second wife (step-mother to John Arthur). Barnabus Arthur s mother was Catharine McDaniel. This surname is common among the Shawnee. Any documented Catharine McDaniels are given a husband not named Arthur (e.g., instead Ward). There was a Gabriel Arthur who was captured by the Cherokee and resided with the Shawnee in the 1600s. He apparently had a family by a Cherokee woman but disappears from the record (not unexpected considering there was no European presence in the Appalachians at that time). On the maternal side, Margaret s mother s direct line maternal ancestor was one Rebecca Wright born circa 1812 in TN. There is a Rebecca Wright of this age from TN, but the Internet insists that she became wife to someone other than her known husband, Joseph Collins, born about 1805 in GA. GENETIC EVIDENCE Early Autosomal Ancestry Tests: In the early days of genetic testing one of the first biogeographical autosomal tests to emerge on the scene was the DNAPrint test offered by DNAPrint Genomics. This measure used 172 single nucleotide markers (SNPs) which were allegedly ancestral informative markers (AIMs). These markers were scattered about the genome, but of course only represented the examination of a single point to a small handful on each of the 22 autosomes. So at best is was a bit chancy as to whether the segment or segments for a potential Native American ancestor would even show up since they might not be targeted at all. The test purported to measure percentage contribution to the genome from each of four continental population sources, European, East Asian, Native American, and African. What follows is the output chart for Flossie Fern. 2

3 The red dot is your most likely outcome from the DNAPrint test. Between the red dot and the gray line are your values that are possible but are 2 times less likely than the red dot. Between the gray line and the blue line are your values that are possible but are 5 times less likely than the red dot. Between the blue line and the yellow line are your values that are possible but are 10 times less likely than the red dot. RESULT TABLE Ancestry Estimate EUROPEAN 90% NATIVE-AMERICAN 10% EAST-ASIAN 0% AFRICAN 0% 3

4 These results seem very clear. The confidence bands are also tightly wrapped around the most likely red dot. So does Flossie Fern have 10% Native American ancestry? The answer is maybe. Over time the test came under closer scrutiny and some of the results simply made no sense. For examples it was not uncommon for Greeks to score 10% and Pakistanis 30% Native American. So while Flossie Fern s results being set on a background of Colonial American, largely or solely Northwestern Europe, the results could be right on but some Colonial Americans with clearly no evidence of any non European ancestry have for example obtained scores of 20% East Asian and 17% Native American. So probably the way to approach these results is to keep them in mind, and blend them with other data only if there is some consistency in the findings. Genome Wide Scans: Since about 2007 the field has improved to the point where using chip technology for medical and ancestry testing has become routine. Commercial testing uses the Affymetrix or Illumina platforms to test over a half a million SNP markers across the entire genome. There are a few regions such as the centromere of each chromosome where there are SNP deserts but generally there are sufficient SNPs such that if a Native American segment of say 10 Mb in length (or 10 cm in genetic distance) is present it can be detected. The entire human genome is about 3 billion base pairs so a segment of a few Mb is very small in relation to the total. Various commercial enterprises have entered this field, but at the time of writing only two are reasonably priced and focus on ancestry results. These are the Family Finder test of Family Tree DNA, and the various features available via 23andMe. One approach to exploring the data for indications of Native American ancestry is to look at the results of the Ancestry Painting feature. An algorithm searches the raw data for each autosome to identify European (reference sample Americans from Utah), African (reference sample the Yoruba tribe from Nigeria), and Asian (Han Chinese and Japanese). These populations come from the original HapMap Project. Native American must be inferred here via the use of the Asian groups as proxy. The empirical finding is that Native Americans tend to have 75% Asian and 25% European plus or minus 10% according to the fact sheet of 23andMe on their Native American Finder feature. What follows is Margaret s finding on each of the two above features. 4

5 As can be observed, Margaret has a small Asian segment at the q end (telomere) of chromosome 4. This may be Asian, Native American or noise. In order to see the specific start and stop points for this segment a program called FireBug was used. The findings are as follows. 5

6 \"4\": [[63508, , \"CC\"], [ , , \"AC\"]], Key: CC=European; AC=Half European Half Asian The Native American Finder feature only assesses Amerind ancestry to 5 generations past. However it will also provide a more precise percentage of the Asian component. Margaret s data is shown below. Here the percentage is given as 0.13% of the entire genome. This would be consistent with having a great great great or further back grandparent, but unlikely to reflect the presence of an Amerind ancestor closer than this. While this finding, in conjunction with the DNAPrint data for Margaret s mother Flossie Fern is consistent with Native American ancestry, it is likely that the purported NA ancestor accounts for 1/64 or less of the total. None the less, the evidence is mounting for a Native American ancestor born before the mid 1800s. Once again this test has come under intense scrutiny and has found to be lacking in that there is a very high probability of false negatives. The reality is that using the Han Chinese to represent segments whose origin is found in the Native Americans from the region east of the Mississippi is bound to result in failing to observe valid Amerind segments, and instead categorizing them in the fall back mode, in other words European. Genetic genealogists (and some population geneticists) have recognized this problem and the former have been in the vanguard of developing more sensitive and sophisticated tests to explore the genome for minority ancestry. Here we will focus on the work of Dr. Doug McDonald and DavidW of the Eurogenes Project. AFTERMARKET BIOGEOGRAPHICAL ANCESTRY TEST PRODUCTS A) Dr. McDonald s Three Biogeographical Ancestry Tests: 1) Dr. McDonald s Whole Genome Analysis The earliest test developed by Dr. McDonald, using the raw data from 23andMe, provides a general breakdown by broad geographical groupings based on the 52 populations (reference samples) in the CEPH-HGDP Panel. Here follow Margaret s results. 6

7 On the quantitative tests, here are results: Europe 98.7% Oceania 0.1% America 2.2% Africa 0.0% E. Asia -1.2% Europe 94.6% S. Asia 5.6% Oceania -0.2% America 1.9% Africa -0.3% E. Asia -1.8% Europe 90.3% Mideast 6.1% S. Asia 4.5% Oceania -0.2% America 2.0% Africa -1.0% E. Asia -1.8% On the bottom one she is unprecedentedly low on S. Asian, and a tad high on Mideast. Pure northwest Europeans are 4.5 Mideast and 8 S. Asian. She is also low on S. Asian on the middle panel. I don't quite know what to make of this. The low S. Asian is telling us something, I don't know what. Brazilians, with a few percent American, and one pure Portuguese show this, also Canary Islands (13 October 2009). Margaret consistently shows scores in the 2% plus Native American no matter what groups are entered into the mix. The consistency is meaningful in that it argues for a valid result although the noise level here is about 1%. 2) Dr. McDonald s All-Genome Assignment Test Here Margaret is right on the borderline and shows 0.4% on the "absolute best fit" group which is considered inside the noise (20 October 2010). In other words her data show a possibility of Native American, but not clearly so. 3) Dr. McDonald s Per-Chromosome Segment Test While the above test is helpful in a global sense of giving a sense of some minority Native American ancestry, it does not provide specific information as to where in the genome that this ancestral component is located. In other words, which chromosomes, the length of the segments, and start and stop locations can be explored with this type of 7

8 testing which is in a broad sense similar to that of the 23andMe Ancestry Painting, except Dr. McDonald uses actual Native American reference samples such as the Maya, Pima, and various South American tribes from the above panel. Recently he has added the Canadian Athabaskans (Na-Dene), the Aleut of Alaska, and the Eastern Greenlanders and Western Greenlanders (Inuit). Also included now are groups in Siberia closely related to the Native North Americans (largely due to apparent back migration) such as the Chukchi (although they are not employed as Native American reference samples here). Here follows Margaret s segment plot analogous to the Ancestry Painting above. Dr. McDonald added, On the per-chrosome test, (she is) just on the edge of significance, at the same sort to level. MArgaret's most likely segment is on the small arm of chromosome 9 right up against the centromere (20 October 2010). 8

9 This segment appears to be about 50 Mb in length, although the centromere contains very little ancestral information hence the focus needs to be on the segments on each side of the dark brown area (particularly the one on the left side of the centromere above). B) Eurogenes BGA101 Project of DavidW: 1) Admixture Analyses using Programs such as STRUCTURE to Give Global Percentages Over the latter months of 2010, David W sent participants a number of analyses from his various admixture analyses. The most recent results were sent on 26 October They include a K=12 mega analysis involving 12 reference populations. Here, the groups are broken down into even more meaningful units such as dividing Amerind into the Central American and South American reference groups. Here the data seems even tighter with for example Europeans such as the French showing little to no percentage factor on the four prime groups of interest to an exploration of Native American ancestry including, Central American, South American, North Asian and East Asian. The following table summarizes the results, with CA4 being David Faux, CA5 his distant cousin with the same Mohawk ancestor, CA8 being a second distant cousin with ties to this ancestor on both her mother and father s side. CA10 is a Canadian with UK and Colonial North American roots but no evidence of any Amerind. These are compared and contrasted to US72: Central Am South Am North Asian East Asian CA4 0.00% 0.61% 1.03% 0.00% CA5 0.72% 0.88% 0.00% 0.00% CA8 0.29% 0.18% 0.00% 0.07% CA % 0.00% 0.00% 0.00% US % 0.00% 0.00% 1.06% It appears that the data from the previous more truncated analysis has here parsed CA5 s Amerind into almost equal amounts of Central and South American but adding up to the same 1.6%. CA4 s scores parse into South America and North Asian, CA8 into the two Amerind groups plus East Asian. Margaret (US72) has a high Central American percentage, and an almost equal component of East Asian. The differences in these results are not altogether clear and reflect the difficulties and sometimes inconsistency of results at the lower percentage levels. The author of this analysis considers this K=12 analysis to be robust (although the percentage of noise will be difficult to determine at lower percentage levels). 2) First RHH Mosaic Chart The first effort of DavidW in creating a chart based on the study of McGinnis et. al. (2010) provided a graphic presentation of the SNP (single nucleotide polymorphism) markers (from the array of over half a million), and a list of these HET (heterozygous) ancestral informative markers. These markers are those which are extremely rare in those 9

10 of European ancestry. They can be seen as non-european markers. More information about the technique is available on David s Eurogenes blog. The chart follows. Here the majority of HETs are those affiliated with Africans. However a careful inspection of each HET resulted in the identification of those found only in Native Americans. The author contacted DavidW and requested that he check one particular area of interest, as well as the chromosome 9 centromere segment identified as possibly Native American by Dr. McDonald. Here follow the PC-MDS charts sent by DavidW. The first two pertain to a region of chromosome 7. 3) PC-MDS Plots for First Mosaic For further information on PC-MDS plots see the above Eurogenes blog. The meaning of the term is principal components multidimensional scaling, to illustrate the clustering 10

11 with a large number of reference groups on the segment of interest. If someone has say a Native American segment from their mother and a European strand (segment) from their father, then the individual will cluster between the two ancestral groupings. Margaret s icon, US72 if found above 0.05 on the X axis. It does appear that if the European strand from her father was removed, the icon would be drawn back to the region where the Athabaskan s cluster. This is a hypothesis that could be confirmed if the segment was phased or parsed into maternal and paternal strands so presumably the former would end up among the Native Americans at the bottom left, and the latter would be found toward the middle or back of the European cluster on the right. 11

12 Above is another view. US72 is found at the 0 point on the Y axis. It appears that this segment is a little confused. It does appear that the clustering is between the Europeans and the Native Americans (as expected with one sttrand that is Amerind), although the scattering is somewhat messy. This is undoubtedly because the author made a mistake and asked DavidW to profile the whole centromere area, whereas according to Dr. McDonald, it is only the area on the left that shows indications of being NA. 12

13 The above close up of the clustering shows that Margaret is positioned to the left of all Europeans, and that Native Americans such as Mayan are cluster mates, there are also East Asians such as the Buryat (a Central Asian group likely closely related genetically to Native Americans). Again, the graph would likely be clearer if the correct start and stop locations had been flagged by the author. DavidW s analysis of the data is that, I think your wife's segments are both potential sites of admixture and worthy of further investigation, especially the chromosome 9 hit when viewed at higher dimensions. But they don't look like 100% cases based on this data. It could go either way with more markers and phased chromosomes (1 December 2010). 4) HET Mosaic Based on Specific Native American Related Markers DavidW has created a Native American specific mosaic by selecting markers that have a clear link only to Native Americans. Margaret s Amerind marker mosaic follows. 13

14 A simple inspection of the above mosaic would suggest a number of small Native American segments for example near the top of chromosomes 7 and 9. David also the list of HETs. It is interesting to note that Margaret obtained 138 HETs compared to 95 for the present author who has one Native American ancestor born A distant cousin of the latter who is two generations closer to the same Amerind ancestor has 125 HETs. Thus Margaret appears to have a larger percentage of Native American input if one can use numbers of HETs as a signal of percentage NA (which has yet to be demonstrated). Within the next week DavidW will be scanning each participant s genome using a sliding window approach where small groups of SNPs will be examined as to ancestry, and then the algorithm will continue using this approach until the whole chromosome is explored for Amerind segments. A MC-PCA plot will be created for each so identified segment. This approach is very advanced and makes a questionable manual pattern recognition inspection approach unnecessary. The results will be included here when available. David K. Faux 8 January